Seasonal and interannual variability of sea surface temperatures in the tropical Pacific, 1969–1991

Deep-SeaResearchl, Vol. 40, No. l 1/12,pp. 2217-2228, 1993.

09674)637/93 $6.09+ 0.00 t~) 1993PergamonPressLtd

Printed in Great Britain.

Seasonal and interannual variability of sea surface temperatures in the tropical Pacific, 1969-1991 THIERRY DELCROIX*

(Received 7 December 1992; in revised form 29 April 1993; accepted 29 April 1993) Abstract--Water sample data derived from a ship-of-opportunity program are used to describe changes in sea surface temperature (SST) for the tropical Pacific. Time series up to 23 y long (19691991) allow investigation of both seasonal and interannual SST variability along four selected shipping tracks crossing the equator near 155°E, and 160, 140 and 100°W. Poleward of 6-8 ° latitude, the seasonal variability increases, reflecting the growing impact of surface thermal forcing. Within 6-8 ° latitude, the seasonal and interannual variabilities in the eastern Pacific are of the same order of magnitude, whereas in the western Pacific the interannual variability is 2-10 times greater than the seasonal variability. Maximum interannual variability appears trapped at and along the equator, in relation with the occurrence of E1Nifio (1976, 1982-1983, 1987 and 1991) and La Nifia (1988-1989) events. Complementing various other local investigations, the present study enables basin-scale SST changes to be documented in a concise fashion.

INTRODUCTION

SEA surface temperature (SST) is probably the most complex physical oceanic parameter. Observing, understanding and ultimately predicting its variability is of major importance because SST anomalies are related to several aspects of the global climate (e.g. PAN and OORT, 1983; PALMERand MANSFIELD, 1984). Anomalies in SST are governed by various processes, which not only vary at different time and space scales, but are also closely dependent on each other: changes in SST involve multiple-scale coupled oceanatmosphere interactions. Basically, SST is modified by two types of processes: those acting through the sea surface and those acting within the water mass. These processes include: (i) the net surface ftux of solar energy, latent and sensible heat, and infrared radiation, and (ii) the horizontal and vertical advections and mixing by various scales of motion. In extra-tropical regions, the surface thermal forcing is thought to control low-frequency SST changes, in contrast with tropical regions where the role of ocean dynamics cannot be dismissed. In the tropics, furthermore, the ocean-atmosphere coupling is enhanced, because the surface thermal forcing is influenced by atmospheric parameters (cloud, wind speed, air temperature, humidity, etc.), all of which are related to SST, and vice versa. Quantifying the relative contribution of processes that cause SST anomalies is thus a fascinating challenge; it underlies the international 1985-1994 Tropical Ocean and Global Atmosphere (TOGA) program (WORLD CLIMATERESEARCHPROGRAM, 1985). * lnstitut Franqais de Recherche Scientifique pour le D6veloppement en Coop6ration (ORSTOM), Groupe S U R T R O P A C - - C e n t r e O R S T O M de Noum6a, BP A5 Noum6a, New Caledonia. 2217

2218

T. DELCROIX

What do we know about SST changes at seasonal and interannual time scales, in the tropical Pacific? With the large body of scientific literature dealing with SST, one might be tempted to assume this parameter to be fairly familiar to us. However, papers include comprehensive descriptions based on atlases and real-time SST products (e.g. LEWTUS, 1982; REYNOLDS,1988), regional descriptions relying on different types of instrumentation (XBT, moorings, drifting buoys, etc.) for various time periods (e.g. WHITE et al., 1985; HAYESetal., 1991; McPHADENetal., 1992), and attempts to model and understand the SST climatology or its variability (e.g. SEA~ERet al., 1988; SEAGER, 1989). Consequently, with information about SST stemming from such a wide variety of sources, instruments, locations and time periods, it is often difficult to pinpoint differences and similarities between findings and to obtain a synthetic view. The present approach complements other analyses by concentrating on basin-scale SST changes measured with the same technique, over a 23 y period. Specifically, this study is an investigation into the seasonal and interannual SST variations, from 1969 to 1991, as inferred from sea water samples collected along merchant ship routes. Obviously, some of the results will be familiar to the reader. However, as noted above, it is useful to document SST variations in a concise fashion. This is expected to highlight basin-scale SST features at seasonal and interannual time scales, and to facilitate comparisons with oceanatmosphere model results. The paper is organized as follows. Section 2 describes the measurements, the resulting data set and data processing. In section 3, emphasis is placed upon the mean and seasonal variability of SST. Then, in section 4, we focus on the SST signature of several El Nifio (1976, 1982-1983, 1987 and 1991) and La Nifia (1988-1989) events that strongly modified the physical environments throughout the Pacific basin. THE DATA The bulk of this work is based on an ongoing ship-of-opportunity program (SOP) developed and operated jointly by the ORSTOM centres in Noum6a (New Caledonia), and Papeete (French Polynesia) since 1969 and 1976, respectively. Among the various parameters gathered by this program, SST measurements are obtained from sea water samples in buckets collected along trans-Pacific routes every 55-110 km (about 25% of the 1969-1972 SST data came from intake measurements but this will not affect the main conclusions). The buckets are hauled up on to the bridge, and SST measurements are performed by volunteer ship officers by means of a Maurer-Amarell thermometer fitted inside the bucket (manufactured by Eurolabo, in Behren l~s Forbach, France). The graduation interval for the thermometer is 1°C; one quarter of a degree is probably the minimum that can be estimated. Out of all the SST measurements available for the tropical Pacific, I have selected about 139,000 observations along four well-sampled shipping tracks, during the 1969-1991 period (Fig. 1). For brevity, the tracks will be referred to as the western track, from 166°E28°S to 142°E-28°N, the central track, from 166°E-28°S to 126°W-28°N, the east-central track, from 152°W-28°S to 126°W-28°N, and the eastern track, from 180°-28°S to 80°W7°N. The ORSTOM-SOP measurements were routinely checked before being entered into the data base. The measurements were edited, checked for gross errors, internal consistency of the data pertaining to each voyage, and climatic limits. Additional

2219

Variability of sea surface temperatures

30°N

20°N

IO°N W r7

L

0°

_..i

10°S

20°S

30"S

130°E

180"

130°W

80°W

LONGITUDE

Fig. 1. Spatialdistribution of sea surface temperature data collected along four mean shipping tracks in the tropical Pacificover the period 1969-1991 period. From west to east, these tracks arc referred to as the western track (W), the central track (C), the east-central track (EC), and the eastern track (E) throughout the paper. validation tests were applied to the selected SST values. Along each track, spurious SST m e a s u r e m e n t s were detected through objective criteria based on multiples (_+5, +4, and then +3.5) of sample standard deviations computed in 2 ° latitude bands. This procedure resulted in the loss of about 1% of the SST observations. In what follows, the time-latitude distribution of the validated SST measurements (Fig. 2) should be kept in mind. Results concerning variability in regions where the n u m b e r of observations is small should be viewed with caution. Note that all the m e a s u r e m e n t s to be analysed end in D e c e m b e r 1991, whereas they start at different times: in August 1969, August 1975, May 1976 and S e p t e m b e r 1974, for the western (56,900 observations), central (38,600), east-central (t3,000) and eastern (30,900) tracks, respectively. As shown in Fig. 2, the along-track SST data are irregularly distributed in latitude and time; they were therefore interpolated onto regular one month by 2 ° latitude (e.g. I°N 1°S) grids by an objective interpolation scheme (Laplacian method). On average, a grid element represents 8.2, 6.7, 3.1 and 8.1 measurements, for the western, central, eastcentral and eastern tracks, respectively. The gridded SST values were then smoothed (Laplacian smoothing), and a 1/4, 1/2, 1/4 filter was applied on each time series. This processing procedure is identical to the one used by DELCROIX and H~NIN (1991) in a companion p a p e r dealing with sea surface salinity data. Estimation of the errors induced by the use of bucket m e a s u r e m e n t s already has been amply d o c u m e n t e d (cf. REVERDIN et al., 1988, and its references). Comparisons between

2220

T. DELCROIX WESTERN T R A C K

CENTRALTRACK

28 N

,6 N

i:i;:i~-!!i,~i~!~

uJ E3 F-

<,

12S 16S 20 S 24 S

~8s

~ 69

, 71

73

, 75

W',~'I",~, 77

79

81

EAST-CENTRAL

uJ £3

<,

83

85

87

89

28 N 24 N 20 N 16N 12N 8N 4N EQ 4S 8S 12S 16S 20 S 24 S 28 S

91

69

71

73

75

TRACK

77

79

28 N 24 N 20 N 16N 12N 8N

4N

=i

. ,

j

75

77

79

81

YEARS

Fig. 2.

83

85

87

r

89

91

87

89

91

i'i,

~.~., = '{#

;

"

; :'~

--

12S 16S 20 S 24 S 28 S 73

85

+~'i . . . . . . .

"

4s

71

83

EASTERN T R A C K

28 N 24 N 20 N 16N 12N 8N 4N EQ 4S 8S 12S 16S 20 S 24 S 28 S 69

81

. !?;~Ni~ !~

U[# : i

69

71

73

75

77

79

81

83

85

87

89

91

YEARS

Time-latitude distribution of sea surface temperature data for the four selectcd tracks.

bucket and other in situ SST measurements are difficult to interpret as the data are rarely simultaneous in space and time. An error estimate was made by comparing bucket SST, collected so as to approximate SOP situation, with simultaneous CTD measurements obtained during O R S T O M - S U R T R O P A C research cruises in the western tropical Pacific. For the 173 comparisons made during 1989-1992, the mean (bucket minus CTD) and rms differences between bucket and CTD measurements obtained from these cruises were - 0 . 1 5 ° C and 0.37°C, respectively. Given that CTD is a very accurate instrument, this provides an order of magnitude of the expected bias and accuracy of a bucket SST. However, precision in a gridded SST value depends on the multiple processing stages applied to the raw data. On a mean, a gridded SST value represents 6.5 observations collected during a given month, within a 2 ° latitude by 10-20 ° longitude band. Although the rms error is reduced roughly by a factor of 6.51/2, inadequate sampling in a month and a 2 ° latitude band, and the effect of grouping the scattered SST data into a mean track, may still alter precision in a gridded value. For all these reasons, the overall accuracy of the gridded SST fields is probably not better than 0.3°C. ANNUAL MEAN AND SEASONAL VARIABILITY As noted by MEYERS (1982), by excluding the E N S O years in the tropical Pacific thermal structure, it is possible to construct means that fall into Gaussian distributions. The longterm annual mean of SST is thus calculated from the monthly gridded values, excluding the 1972-1973, 1976, 1982-1983 and 1987 E N S O years as defined by QUINN et al. (1987).

Variability of sea surface temperatures 30

ANNUAL MEAN

J .

28

-"

26

SEASONAL VARIABILITY

"[

" ' 1 ' ' ' 1 ' ' ' ' ' '

' 1 ' ' ' 1 ' ' '

'

/

---

\~,

/

E /"

\ \

,,/

24

c

'"

~

2

. , , I , . , I , , , I ,

i . . . i , , .

2221

E"

//

.._

22

20

-24-

' :16

'-8

0

' 0 ' " '8

" " "16

" 24

I

,

,

,

I

,

,

,

I

.

.

.

.

.

.

.

I

,

,

,

I

,

,

,

i

10

,

© EC /" Vx / J,"'.,, ~

:2~, ' :~k

\

-~ . . . . . . LATITUDE

'

~

"

" ._,,'

~

I

.

.

.

.

.

.

.

.

-16

.

-8

i

T

I

8

16

24

INTER. / SEASONAL VAR.

INTERANNUAL VARIABILITY _,

'

-24

,,

,//C 2~,

,

J

,

,

,

I

,

.

t

.

.

.

.

.

8 ®

.

1

~ . . .

~

.

,

~

/l/vv

6

/

o/,?,, / -24

-16

-8

8

16

24

LATITUDE

Fig. 3. (a) Annual mean (°C), (b) seasonal variability (°C), (c) interannual variability (°C), and (d) interannual vs seasonal variabilityof sea surface temperature for the western (W), central (C), east-central (EC) and eastern (E) tracks.

Similarly, a mean year is constructed by omitting the E N S O years; the standard deviation (o~) of the 12 mean months is used as a measure of the seasonal variability. The long-term annual means of SST are presented in Fig. 3a. Maximum SST values (>29°C) are located along the western track, between 10°S and 8°N, decreasing poleward and toward the east. The SST warmer than 28°C, temperature threshold frequently used to delimit the warm pool region, is found along the western, central, and east-central tracks, within 14°S-18°N, 14°S-2°N, and 12-8°S, respectively. Except along the western track, local SST maxima appear on either side of, but asymmetrically about, the equator. The SST maxima at 6°N (155, 137 and 85°W), 8°S (170°W), 12°S (145°W) and 14°S (135°W) seem to be associated with the zonally-oriented Intertropical Convergence Zone (ITCZ) and the NW/SE-oriented South Pacific Convergence Zone (SPCZ). The way I T C Z and SPCZ are coupled to the underlying SST, however, is still an open question. Between the local maxima, a minimum SST lies on the equator, characterizing the progressive appearance of the equatorial upwelling to the east. These annual means in SST agree closely with the results of HOREL (1982) and LEVmJS (1982). The amplitude (os) of the seasonal variability in SST increases poleward in each hemisphere from 8-10 ° latitude (Fig. 3b). The largest values are obtained for grids in the western Pacific, and in particular in the Kuroshio region (Os=2.8°C at 28°N). In the equatorial band, the seasonal variability is very slight (c& < 0.25°C) for the two westernmost tracks and increases toward the east to reach 0.5°C at 140°W and 1.5°C at 100°W. This difference reflects the eastward increase of the seasonal signal in the equatorial upwelling

2222

uJ £3

<,

T. DELCROIX SST : WESTERN TRACK

28 N 24N 20 N 16N 12N 8N 4N EQ 4S 8S 12S 16S 20 S 24 S 28 S

24 N 20 N 16N 12N 8N 4N

D

uJ £3

5

SST : CENTRAL TRACK

28N

J

F

M

A

M

J

J

A

S

O

N

D

J

SST : EAST-CENTRAL TRACK

28 N 24 N 20 N 16N 12N 8N 4N EQ 4S 8S 12S 16S 20 S 24 S 28 S

~~

'

~

~

D

J

F

M

A

M

~

J.

J

....

A

S

O

N

D

J

0

N

D

J

SST : EASTERN TRACK

28 N 24N 20 N 16N 12N 8N

EQ 4S 12 S 16S 2O S 24 S D

J

F

M

A

M

J

J

A

S

STANDARD YEAR

0

N

D

J

ir

D

J

F

M

A

M

J

J

A

S

STANDARD YEAR

Fig. 4. Mean seasonal cycle of sea surface temperature for the four selected tracks. Contour intervals are I°C, except for the dashed lines which denote the 29.5°C isotherm. Shaded areas represent sea surface temperature above 28°C,

extending from the date line to the Galapagos Islands. Note, however, that the maximum SST seasonal variability is not located right at the equator but slightly displaced southward, in agreement with the findings of WEAREet al. (1976). As can be expected, away from the equatorial band, the SST is warmest (coldest) during the summer (winter) hemisphere (Fig. 4). This merely reflects the considerable impact of the surface thermal forcing in controlling extra-tropical SST (Liu and GAUTIER, 1990). In the equatorial band, the seasonal SST change is well-marked for the eastern track only. There, the SST is warmest during March-April when the ITCZ is farthest south and the zonal winds at the equator are at their weakest resulting in a minimum-strength upwelling. Conversely, the SST is coldest during September-October with the ITCZ is at ~ts northernmost position, and when there are strong easterly winds (HASTENRATHand LAMB, 1977) and a westward surface current (HALPERN, 1987), yielding maxima in vertical and horizontal advections of cold water. In agreement with HOREL(1982), there is a suggestion on Fig. 4 that the seasonal SST warming moves westward from the equatorial eastern Pacific, the warmest equatorial SST being found in March-April at 100°W, May-June at 140°W, July-August at 160°W and January-February at 155°E. This is closely associated with the westward phase propagation of the seasonal signals in zonal winds (MEVERS, 1979) and zonal currents (McPHADEN and TArt, 1988). Strong interactions between the ocean and the atmosphere are obviously involved in the equatorial waveguide, at seasonal time scale. As shown in Fig. 4, the seasonal SST signal at some latitudes can be easily represented

2223

Variability of sea surface temperatures

PHASE

AMPLITUDE 360

, ~ . ,

,

,

I

,

,

,

I

.

.

.

.

.

.

i

0

8

.

.

~

.

I

.

270

180

90

0 -24

-16

-8

0

8

16

24

-24

-16

% VARIANCE EXPLAINED 100

j

,

,

.

J

80

.

.

.

.

.

\

.

.

.

I

F,

.

.

,

I

....

60

,

,

,

-8

16

24

LATITUDE

I - -

//w

40 20

-24

-16

-8

0

8

16

24

LATITUDE

Fig. 5. Least squares fit to an annual harmonic of sea surface temperature for the western (W), central (C), east-central (EC) and eastern (E) tracks: (a) amplitude (°C), (b) year-day phase of warmest sea surface temperature, and (c) per cent of total variance explained by the fit.

by a periodic function of one year duration. For each track, least-squares fit to an annual harmonic was thus performed to assess how representative the annual SST cycle is of the total SST signal. The amplitude and phase of the annual harmonic, together with the percent of variance explained by the fit (Fig. 5) corroborate the previous discussion. The amplitude increases poleward from about 8° latitude, from east to west, and there is a 180° phase shift between the northern and southern hemispheres. In the equatorial band, the amplitude is negligible for the two westernmost tracks, and it reaches maximum values (2°C) at 2°S along the eastern track. Interestingly, the annual SST phase within 2-4 ° around the equator behaves more like the phase of the southern hemisphere for the western track, and more like the phase of the northern hemisphere for the central track. The reason for this is unclear. It is also noteworthy that there are regions near the equator where the annual SST cycle is not very significant: the annual SST cycle fits less than 20% of the total SST variance west of 140°W. In contrast, the annual cycle is a major signal in the eastern Pacific, accounting for at least 50% of the total variance. Because the annual cycle appears coupled with the interannual E1 Nifio variability (RASMUSSON and CARPENTER, 1982), processes that govern its evolution, although only partly understood (SEAGERet al., 1988), must be precisely quantified. INTERANNUAL

VARIABILITY

In the previous section, we mentioned regions where the annual SST cycle explains only a small part (<20%) of the total SST variance. To ascertain the relative contributions of

2224

T. DELCRO1X

interannual vs seasonal SST changes, we computed anomalies of SST with reference to the 12 mean months; the standard deviation (oi) of these anomalies was used as a measure of the interannual SST variability. Estimates of the amplitude of the seasonal with respect to the interannual signals were obtained by computing the ratio (ai/as) of the standard deviation of the interannual variability (oi) to the standard deviation of the seasonal variability (tTs). The interannual variability of SST (Fig. 3c) is distributed fairly similarly along the three easternmost tracks. Within 10° off the equator, ai can be roughly represented by a Gaussian function with the acme at the equator (~ri = 1-1.5°C). Such a distribution does not conflict with model results indicating that interannual El Nifio-related variability might stem from anomalous equatorially-trapped eastward surface advection associated with long equatorial waves (HARRISON and SCHOPF, 1984). Further poleward, a i increases in regions where the meridional gradient in mean SST is high (i.e. in thermal frontal zones). Along the western track, the interannual variability is almost constant, ranging from 0.5 to 0.75°C. Obviously, high values of the ratio (oi/as) are found in the regions where the seasonal SST signal is weak. As shown in Fig. 3d, notable track to track (i.e. almost zonal) variations of (tyi/os) are found mainly within 8°N-8°S, particularly near the equator where the ratio gradually increases from east to west. This feature clearly illustrates one fundamental difference in SST changes between the western equatorial Pacific, where interannual variations exceed seasonal variations, and the eastern equatorial Pacific, where interannual and seasonal variations are of the same order of magnitude. Having presented a brief description of the interannual SST changes, we shall now examine time series of sea surface temperature anomalies (SSTA), between 20°S and 20°N to highlight the basin-scale features (Fig. 6), and at the equator, to better apprehend the timing of the anomalies (Fig. 7). Bear in mind that "at the equator" signifies the mean for I°N-I°S. Looking at the eastern track in Fig. 6, we see that significant positive SSTA, exceeding 0.5-1°C, appear during five remarkable periods that characterize the 1976 moderate E1 Nifio event, the 1979 aborted event, the 1982-1983 very strong event, the 1987 moderate event, and the premise of the 1991-1992 event (QuINN et al., 1987; DON~UY and DESSIER, 1983; ENFIELD et al., 1992). E1 Nifio-related SSTA also can be traced along both the poorly-sampled east-central track (except in 1976) and the central track, reflecting the basin-scale impact of El Nifio events. In agreement with Fig. 3c, positive SSTA warmer than I°C appear somewhat trapped at the equator, extending on a mean from 8°N-8°S. As noted previously, this is perhaps in relation with the signature of equatorially-trapped surface current anomalies associated with equatorial waves. Only detectable along the eastern track, negative SSTA occur near the end of 1975 at a time of positive Southern Oscillation Index (SOl) (Climate Analysis Center, Washington D.C., U.S.A.). In the second half of 1983, negative SSTA exist near the equator for the east-central and central tracks, in agreement with other data sets (HARRISONet al., 1990). As was to be expected, the strong 1988-1989 La Nifia event results in well-marked negative SSTA ( < - 2 ° C ) for the three easternmost tracks. (Note the lack of data between March and October 1988 along the eastern track.) Because SSTA are generally maximal at the equator, their timing and amplitude can be more easily traced using Fig. 7. Note that from west to east, the selected tracks cross the equator near 155°E and 160, 140 and 100°W. The 1976 moderately strong E1 Nifio event results in above normal equatorial SSTA from April 1976 to March 1977 near 100°W, and

2225

Variability of sea surface t e m p e r a t u r e s

SSTA: CENTRAL TRACK • v ~ :"!'~ !it":' ~i,',', !i

SSTA: WESTERN TRACK 20 N 16N 12N uJ £3

i!I![

8N 4N EQ

5

4S

8S

8S

il

'

12S 16S 20 S 75

77

79

SSTA 20 N

~.~,,

16N

~"~

12N

uJ

£3

83

85

87

89

75

91

,~i~i:I::~,,,,~t~i .,i~i!i!-!"~i'"~

i~i!'O

~c!J

79

81

83

85

87

89

91

89

91

20 N 16N 12N

"':.MI'>~

:

77

SSTA: EASTERN TRACK

8N

8N

4N

4N EQ

<,

81

EAST-CENTRAL TRACK

i:,,,, ".~

~J!

4S 8S

EQ

.3 St~..', ' ",~

~~ '~'-,,::i~t,}o'7!)iJ,~~,°i"t

4S 8S

12S

12S 16S

16S

20 S

20 S 75

77

79

81

83

85

87

89

75

91

77

79

81

83

85

87

YEARS

YEARS

Fig. 6. T i m e - l a t i t u d e d i a g r a m s of sea surface t e m p e r a t u r e a n o m a l i e s relative to the m e a n seasonal cycles, C o n t o u r intervals are I°C, except for the _+0.5°C anomalies. T h e zero c o n t o u r is d r o p p e d for clarity, and the d a s h e d lines d e n o t e negative anomalies. S h a d e d areas r e p r e s e n t sea surface t e m p e r a t u r e a n o m a l i e s a b o v e 0.5°C.

2

I

I

I

I

I

I

I

1

I

I

I

1 _1

0 --2

Fig. 7.

I

I

//~,, 75 I

i771

1791

i 81 I

183 i

185 I

187 I

189 ~

I

[

• 191

E q u a t o r i a l sea surface t e m p e r a t u r e a n o m a l i e s (°C) relative to the m e a n seasonal cycles, for the western (W), central (C), east-central (EC) and eastern (E) tracks.

2226

T. DELCRO1X

from April 1976 to early 1978 near 160°W. (Note that the lack of the El Nifio signal near 140°W may reflect the poor data coverage along the east-central track.) Maximum equatorial SSTA are reached in August 1976 near 100°W (2.7°C), and in January 1977 near 160°W (1.5°C). This is in qualitative agreement with the SSTA along the coast off Peru, as reported in WYRTKI(1979). DONGUV and DESSIER (1983) have reported the occurrence of a very weak E1 Nifio during 1979, from sea surface salinity and temperature data collected along the same tracks as in Fig. 1. Figure 7 shows that positive SSTA occur during some months in 1979 and 1980 near 100,140 and 160°W. It should be noted that three month periods with SSTA warmer than I°C only appear along the poorly-sampled east-central track. SSTA derived from operational products (Climate Analysis Center, Washington, D.C., U.S.A.) in NliqO 1+2 (0-10°S; 80-90°W), NllqO 3 (6°N-6°S; 90-150°W), and NIlqO 4 (6°N-6°S; 150°W-160°E) regions never exceed 1°C during 1979-1980. In the western equatorial Pacific, prior to the very strong 1982-1983 event, there is a trend for SSTA to increase from 1975 to early 1982 (about 0.25°C y-l) (Fig. 7). There, SSTA warmer than I°C only happen in 1981 and early 1982, which is consistent with the results reported by MEYERSel al. (1986), based on XBT data. Whether this warming was significant and/or instrumental in setting the stage for the strongest event on record is not clear. Positive equatorial SSTA are observed during the first quarter of 1982, almost simultaneously near 100, 140 and 160°W. There is no clear evidence, with the present data set, that SSTA first occur in the western equatorial Pacific and subsequently propagate eastward, as suggested by GILL and RASMUSSON (1983). The warmest equatorial SSTA occur in both January (4°C) and June (4.4°C) 1983 near 100°W, and in January 1983 near 140°W (3.9°C) and 160°W (2.3°C). Interestingly, the SSTA near 100°W present two distinct peaks (January and June 1983), in qualitative agreement with the E1 Nifio composite of RASMUSSONand CARPENTER(1982) except that the present episode started several months earlier and lasted longer than the composite. This double peak structure may reflect the signature of first-baroclinic mode Kelvin waves (CANE, 1984): the first peak is primarily due to the changes in the winds close to the dateline, the second peak results from the collapse of the trade winds in the central Pacific. However, the cause and effect relations of Kelvin wave-induced anomalous advection and warm SST are still insufficiently understood. From mid-1983, the equatorial SST decreases, returning to normal in October 1983 near 100°W, and in July 1983 near 140 and 160°W. The warm 1982-1983 episode has ended, and the eastern half of the equatorial Pacific becomes colder than normal in 1984 and part of 1985. All these findings agree reasonably well with the largescale pattern of SST evolution in 1982-1983, in particular with near-surface ll0°W mooring data (HALPERN,1987) and with XBT measurements along the central track (TouRNIER, 1989). During the 1986-1987 El Nifio event, the positive equatorial SSTA begins in mid-1986 near 100, I40 and 160°W and remains above normal throughout 1987 (Fig. 7). The largest positive equatorial SSTA occur both in May (2.3°C) and September (2.2°C) 1987 near 100°W, and in June 1987 near 140°W (1.7°C) and 160°W (1.8°C). As during a canonical E1 Nifio, there is a double peak in SSTA structure in the east, although variations between May and September 1987 approach the measurement accuracy. Equatorial SSTA remains above normal, decreasing from mid-1987 to January 1988. Then from February to the end of 1988, equatorial SSTA drop by 3-4°C near 140 and 160°W (no data near 100°W) as the 1986-1987 El Nifio warm event give way to a cold La Nifia lasting to the end of 1989.

Variability of sea surface temperatures

2227

D e s p i t e q u a n t i t a t i v e d i s a g r e e m e n t s o f t h e o r d e r o f 1°C, o u r t i m e e v o l u t i o n o f e q u a t o r i a l S S T A d u r i n g t h e 1 9 8 6 - 1 9 8 8 p e r i o d is s u p p o r t e d b y MCPHADEN a n d HAYES (1990), w h o u s e d e q u a t o r i a l m o o r i n g S S T d a t a at 110 a n d 140°W. Q u a n t i t a t i v e d i s a g r e e m e n t s m a y reflect the different type of measurements, the use of different climatologies, and the fact t h a t t h e p r e s e n t S S T A r e p r e s e n t m e a n s o v e r 2 ° l a t i t u d e by 1 0 - 2 0 ° l o n g i t u d e b o x e s n o t strictly c e n t r e d a r o u n d m o o r i n g l o c a t i o n s . F i n a l l y , e q u a t o r i a l S S T A w a r m e r t h a n 0 . 5 - 1 ° C h a p p e n in m i d - 1 9 9 1 n e a r 160, 140 a n d 100°W: t h e y e v i d e n c e t h e e a r l y e f f e c t s o f t h e 1 9 9 1 - 1 9 9 2 E! Nifio. T h e b e g i n n i n g o f t h e e v e n t o n l y is c a p t u r e d h e r e ; S S T A s e e m in a g r e e m e n t w i t h o p e r a t i o n a l p r o d u c t s ( C l i m a t e A n a l y s i s C e n t e r , W a s h i n g t o n , D . C . , U . S . A . ) . Full d e s c r i p t i o n s o f t h e e v e n t s h o u l d b e available soon.

Acknowledgements--Numerous officers and crews of merchant and navy ships voluntarily participate in the ORSTOM ship-of-opportunity programme, collecting sea surface temperature measurements among other parameters. The present analysis represents the combined efforts of many ORSTOM colleagues, and in particular owes much to J. R. Donguy and C. H6nin in initiating and continuing the surface ship-of-opportunity programme. Programming support was mainly provided by F. Masia. All these contributions are gratefully acknowledged.

REFERENCES CANE M. (1984) Modeling sea level during El Nifio. Journal of Physical Oceanography, 14, 1864-1874. DELCROIXT. and C. HI~NIN(1991) Seasonal and interannual variations of sea surface salinity in the tropical Pacific Ocean. Journal of Geophysical Research, 96, 22135-22150, DONGUYJ. R. and A. DESSIER(1983) El Nifio-like events observed in the tropical Pacific. Monthly Weather Review, 111, 2136-2139. ENFIELDD. B., R. CAL1ENESand L. PIZARRO(1992) Oceanographic conditions along the coast of south America during the 1991-92 El Nifio/Southern oscillation. TOGA Notes, 8, 1-4. GILL A. and E. RASMUSSON(1983) The 1982-83 climate anomaly in the equatorial Pacific. Nature, 305,229-234. HALPERN D. (1987) Observations of annual and El Nifio thermal and flow variations at 0-110°W and 0-95°W during 1980-1985. Journal of Geophysical Research, 92, 8197-8212. HARRISOND. E. and P. S. SCHOPF(1984) Kelvin-wave induced anomalous advection and the onset of surface warming in E1 Nifio events. Monthly Weather Review, ll2,913-933. HARRISOND. E., B. S. GIESE and E. S. SARACHIK(1990) Mechanisms of SST change in the equatorial waveguide during the 1982-83 El Nifio. Journal of Climate, 3, 173-188. HASTENRATHS. and P. J. LAMB(1977) Climatic atlas of the tropical Atlantic and eastern Pacific oceans. 97 charts. University of Wisconsin Press, Madison. HAYES S., P. CHANG and M. J. MCPHADEN (1991) Variability of the sea surface temperature in the eastern equatorial Pacific during 1986-1988. Journal of Geophysical Research, 96, 10553-10566. HORELJ. (1982) On the annual cycle of the tropical Pacific atmosphere and ocean. Monthly WeatherReview, 110, 1863-1878. LEVITUSS. (1982) Climatologicalatlas of the world ocean. NOAA Professional Paper, 173 pp, U.S. Govt. Print. Off., Washington D.C. Liu T. and C. GAUTIER(1990) Thermal forcing on the tropical Pacific from satellite data. Journal of Geophysical Research, 95, 13209-13217. MCPHADEN M. J. and S. P. HAYES(1990) Variability in the eastern equatorial Pacific ocean during 1986-1988. Journal of Geophysical Research, 95, 13195-13208. McPHADEN M. J. and B. A. TAFT (1988) Dynamics of seasonal and intraseasonal variability in the eastern equatorial Pacific. Journal of Physical Oceanography, 18, 1713-1732. MCPHADEN M. J., F. BAHR, Y. DuPENHOAT,E. FIRING, S. P. HAYES,P. P. NnLER, P. L. RICHARDSONand J. TOOLE (1992) The response of the western equatorial Pacific ocean to westerly wind burst forcing: November 1989January 1990. Journal of Geophysical Research, 97, 14289-14303.

2228

T. DELCROIX

MEYERS G. (1979) Annual variation in the slope of the 14°C isotherm along the equator in the Pacific ocean. Journal of Physical Oceanography, 9, 885-891. MEYERS G. (1982) Interannual variation in sea level near Truk Island: A bimodal seasonal cycle. Journal of Physical Oceanography, 12, 1161-1168. MEYERSG., J. R. DONGUYand R. K. REED (1986) Evaporative cooling of the western equatorial Pacific Ocean by anomalous winds. Nature, 323, 523-526. PAN Y. H. and A. H. OORT (1983) Global climatic variations connected with sea surface temperature anomalies in the eastern Pacific ocean for the 1958-1973 period. Monthly WeatherReview, 111, 1244-1258. PALMERT. and D. A. MANSFIELD(1984) Response of two atmospheric general circulation models to sea surface temperature anomalies in the tropical east and west Pacific. Nature, 310,483-485. QUINN W. H., V. Z. NEAL and S. E. ANTUNEZ DE MAYOLO(1987) El Nifio occurrences over the past four and a half centuries. Journal of Geophysical Research, 92, 14449-14461. RASMUSSONE. M. and T. H. CARPENTER(1982) Variations in tropical sea surface temperature and surfacc wind field associated with the southern Oscillation/E1 Nifio. Monthly WeatherReview, 110, 354-384. REVERDIN G., Y. DUPENHOATand J. SERVAIN(1988) Sea surface temperature: a comparison betwccn ship reports from marine decks and ship-of-opportunity subsurface data in the tropical Atlantic ocean. Tropical OceanAtmosphere Newsletter, 47, 1-4. REYNOLDS R. W. (1988) A real-time global sea surface temperature analysis, Journal of Climate, 1, 75-86. SEAGER R. (1989) Modeling tropical Pacific sea surface temperature: 1970-1987. Journal of Physical Oceanography, 19, 419-434. SEAGERR., S. E. ZEBIAK and M. A. CANE (1988) A model of the tropical Pacific sea surface temperature climatology. Journal of Geophysical Research, 93, 1265-1280. TOURNIER R. (1989) Variabilit6 de la structure thermique et des courants ~ l'ouest et au centre de l'oc6an Pacifique tropical. Th~se de doctorat, 154 pp, Universit6 de Paris VI, Paris. WEARE B. C., A. R. NAVATO and R. E. NEWELL (1976) Empirical orthogonal analysis of Pacific sea surface temperature. Journal of Physical Oceanography, 6, 671-678. WHITE W. B., G. MEYERS,J. R. DONGUYand S. E. PAZAN (1985) Short-term variability in the thermal structure of the Pacific ocean during 1979-82. Journal of Physical Oceanography, 15,917-935. WORLD CLIMATE RESEARCH PROGRAM (1985) Scientific plan for the Tropical Ocean and Global Atmosphere program, Publ. 3, World Meteorological Organization, Geneva, 146 pp. WYRTKI K. (1979) The response of the sea surface topography to the 1976 El Nifio. Journal of Physical Oceanography, 9, 1223-1231.